Data analytics is often described as turning raw data into actionable insights. This sounds punchy for sure, but the steps it involves are less clearly defined, ranging from tidying and summarizing data to understanding why results present the way they do.
But of all the tools available to data analysts, perhaps the most powerful is prescriptive analytics. That’s because we can use it to explore the relationship between past events and future outcomes, helping determine what actions should be taken. When used effectively, prescriptive analytics is the key to unleashing data’s full potential.
But what exactly is prescriptive analytics, and what does it involve? In this introductory post, we’ll explore this topic in-depth, answering:
- How does prescriptive analytics work?
- How is prescriptive analytics used?
- Advantages of prescriptive analytics
- Disadvantages of prescriptive analytics
- Prescriptive analytics use cases
- Key takeaways
Let’s dive right in.
1. How does prescriptive analytics work?
Background
To grasp the ins and outs of prescriptive analytics, it first helps to know about a few other types of data analytics. Namely, these are descriptive and diagnostic analytics (which identify what has happened and why) and predictive analytics (which explores what might happen in the future).
Where prescriptive analytics comes in
Meanwhile, prescriptive analytics builds on these techniques to determine a suitable course of action based on reasonable forecasts. For instance, if a company’s profits unexpectedly surge or dip, descriptive and diagnostic analytics can help you determine why.
Predictive analytics will help you figure out whether they will continue this trend in the future. You’d then use prescriptive analytics to identify appropriate steps forward, capitalizing on opportunities and mitigating risks.
Warning: Not a crystal ball
Prescriptive analytics relies on big data combined with carefully defined business rules, machine learning algorithms, and other types of computational modeling. With all this power behind it, it’s tempting to think of prescriptive analytics as a crystal ball, providing a single course of action towards a guaranteed outcome.
In reality, this isn’t how it works. Instead, it helps us outline several possible courses of action and the predicted results of each. So it’s not a crystal ball, but it is a powerful tool for decision-makers who want to make more informed choices.
A word of warning, though! Because prescriptive analytics is the most complex type of data analytics to get right, it’s also the most error-prone. It requires a complex combination of technical, communication, and business skills, explaining why there is currently a surge in demand for data analysts with this kind of expertise.
2. How is prescriptive analytics used?
Whether we call it prescriptive analytics or something else, data-driven decision-making is the logical conclusion of all this complex work. You wouldn’t carry out data analytics otherwise! But how, exactly, is prescriptive analytics used to inform decision-making? Some examples include:
- Driving recommendation engines and individualized content curation on company websites, blogs, streaming services, e-commerce, and social media.
- Predicting financial fraud before it occurs and taking steps to prevent it.
- Helping identify the most lucrative investment areas for energy companies (from drilling for gas to investment in green technologies).
- Predicting the effect of pre-emptive medical interventions on patient outcomes.
- Identifying which sales leads are most likely to convert into customers and creating nudges for these particular customers.
Not every organization is on board with this approach yet, though. That’s because prescriptive analytics requires high-level skills in areas like machine learning, which not all organizations can easily access. Add to this the expense of hiring appropriate staff and the broader challenges of creating an effective prescriptive analytics strategy, and we begin to understand the problem.
Only recent advances in technology have made prescriptive analytics more widely available. As new platforms and technological solutions emerge, prescriptive analytics becomes increasingly accessible to small and medium-sized businesses and startups. So while it’s an evolving field, this makes it an exciting one to get involved in.
3. Advantages of prescriptive analytics
Now that we’ve got a grasp on what prescriptive analytics involves and where to apply it, what are its key benefits? Advantages of prescriptive analytics include:
- It outlines a selection of different possible courses of action and presents the predicted outcomes of each.
- Prescriptive modeling and machine learning algorithms work more quickly and accurately than humans, providing insights directly at the point of need.
- Machine learning also removes many concerns around human error (although this has a flip side, which we’ll see below).
- As prescriptive frameworks evolve, future data analysts will build on these approaches to inform best practice.
There’s no doubt that prescriptive analytics is one of the most powerful tools at a data analyst’s disposal. But that does not mean it is foolproof! This leads us to our next section…
4. Disadvantages of prescriptive analytics
As powerful as prescriptive analytics can be, it’s not necessarily the holy grail many take it for. In this section, we highlight some of the potential disadvantages of using prescriptive analytics:
- It is often poorly distinguished from predictive analytics, which makes defining best practice more challenging.
- Overconfidence in machine learning’s predictive capabilities can lead to people being swayed to follow its advice, regardless of whether or not it is correct.
- Similarly, machine learning predictions may sway people to a course of inaction or complacency (potentially just as risky).
- If the underlying data used is incorrect, the output recommendations will be skewed (garbage in, garbage out).
- In cases where decision-making is automated, there’s the risk that an algorithm will take an inappropriate course of action.
- Prescriptive analytics requires close surveillance by highly-qualified analysts with experience in machine learning, which is time-consuming and costly.
These issues require careful management and quality assurance. As long as they are appropriately addressed, prescriptive analytics have plenty of potential for improving decision-making. To illustrate, let’s take a look at some use cases.
5. Prescriptive analytics use cases
So far, we’ve explored what prescriptive analytics is, as well as its strengths and potential weaknesses. Next, let’s look at some real-world use cases.
Maps apps
Online maps like Google Maps (and other navigation tools reliant on GPS) all use prescriptive analytics. Using various data sources—from location to live weather and traffic reports—they calculate and predict travel times between two or more destinations.
Notably, this involves providing various options, along with predicted travel times for each. Do you want to drive, cycle, walk, or take public transport? Each option gives different arrival times based on the route, transport speed, and other data. This is what prescriptive analytics is about—providing several choices accompanied by likely outcomes.
Self-driving cars
Self-driving cars, which rely on the same data as GPS apps, are another example of the potential of prescriptive analytics. Autonomous vehicles use a combination of historical and real-time data to identify landmarks, objects, other cars (both stationary and non-stationary), and pedestrians.
Based on these data, self-driving cars model various courses of action and their outcomes. For instance, if turning left leads to ‘safety’ and turning right means ‘collision’, the car will (in theory!) automatically select the safest route. Using real-time course corrections, you should arrive at your destination via the safest, most efficient route.
Travel industry pricing
If you’ve ever booked a train journey, flight, or hotel room online, you’ll know that price comparison sites are big business. These sites have disrupted the way we book holidays and travel companies have had to adapt. They increasingly use prescriptive analytics to identify past trends and anomalies, automatically adjusting their pricing in accordance to their data.
Tracking customer demand, weather data, major calendar events, and so on, airlines will rise or drop their prices accordingly. For example, if bookings on one day are lower than the same day the previous year, airlines may drop their prices slightly to increase bookings. This sounds straightforward, but it takes place at dizzying speed to keep up with competitor pricing.
Hospitals and healthcare management
Did you know there are many healthcare-specific prescriptive analytics platforms? These allow clinicians and healthcare managers to compare patient data with waiting times and other factors, like the costs associated with different treatments or appointment no-shows. They can then evaluate and suggest the most suitable way forward to keep hospitals running as smoothly as possible.
One example is the issue of hospital readmissions; a costly problem in the sector. By analyzing patient data against external factors like environment and socioeconomic status, clinicians can predict the likelihood of different demographics being readmitted to the hospital once they’ve been discharged. Predicting this before it occurs, they can improve health outreach accordingly, reminding patients to keep up with their diets or take medication, and so on. This can improve patient health, reduce readmissions, and, ultimately, improve the running of the hospital.
6. Key takeaways
So there we have it, your complete introductory guide to prescriptive analytics! In this post, we’ve learned that:
- Prescriptive analytics goes beyond predictive analytics by identifying potential courses of action and their predicted outcomes.
- Prescriptive analytics typically relies on big data, machine learning, and other types of computational modeling.
- As one of the most complex forms of data analytics, predictive analytics calls for a high degree of technical, communication, and business skills.
- Common uses include content recommendation engines, predicting and preventing financial fraud, improving healthcare management, and choosing lucrative business investments.
- The main advantage of prescriptive analytics is that it can be automated using machine learning.
- One disadvantage of prescriptive analytics is the degree of expertise it requires, which is both costly and time-consuming.
As an emerging field heavily reliant on machine learning tools, prescriptive analytics does not come without risks. However, used responsibly, it has compelling potential. If you’re interested in learning more about data analytics careers, check out this free, 5-day data analytics short course or read the following introductory topics: